Enterprise Bookmark Manager — Built to Replace a Paid SaaS

As a Monitoring Team Supervisor, I noticed we were paying for a bookmark management tool (Toby) to organize our hundreds of monitoring dashboards, runbooks, and action links. Instead of renewing the subscription, I built SROBY — a free alternative that runs entirely on Google Apps Script.

💰 Result: Eliminated recurring SaaS costs while adding features the paid tool didn't have (real-time multi-user sync, user attribution, custom categories).

Our monitoring team manages hundreds of critical links:

• 📊 Grafana dashboards (dozens of them)
• 📋 Runbooks & action procedures
• 🔔 Alert management consoles
• 🛠️ Internal tools & configurators
• 📈 Metrics & reporting platforms

25 team members needed instant access to these resources, organized by category, with the ability to quickly search and open multiple links at once.

Custom multi-tier cache implementation mimicking Redis behavior:

// Redis-like API implemented from scratch
redisGet(key)           // O(1) cache retrieval
redisSet(key, val, ttl) // With TTL expiration
redisDel(key)           // Key invalidation  
redisFlush()            // Full cache purge

Three-Tier Cache Hierarchy:

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│  L1: Redis-like │ →  │  L2: Script     │ →  │  L3: Properties │
│  (In-memory)    │    │  Cache Service  │    │  Service        │
│  TTL: 6 hours   │    │  TTL: 6 hours   │    │  Persistent     │
└─────────────────┘    └─────────────────┘    └─────────────────┘
        ↓ MISS              ↓ MISS                   ↓ HIT
    Check L2 →          Check L3 →           Return + Populate L1/L2

Full pub/sub messaging system with enterprise features:

EventBus {
    topics: Map(),              // Topic-based routing
    eventQueue: [],             // Ordered message queue
    deadLetterQueue: [],        // Failed message storage
    maxRetries: 3,              // Retry policy
    retryDelay: 1000,           // Exponential backoff base
    broadcastChannel: BroadcastChannel  // Cross-tab sync
}

Features:

• 📬 Topic-based Pub/Sub — Decoupled component communication
• ⚖️ Priority Subscriptions — Handlers execute by priority order
• 🔄 Exponential Backoff Retry — 1s → 2s → 4s automatic retry
• 💀 Dead Letter Queue (DLQ) — Failed events preserved for debugging
• 📡 BroadcastChannel API — Cross-tab real-time synchronization
• 💾 Event Persistence — Queue survives page refresh via localStorage

Distributed actor pattern with state partitioning:

VirtualActor {
    instanceId: "actor_1706961234567_a1b2c3d4",
    partitionId: 2,           // Consistent hash assignment
    localState: Map(),        // Actor-local state
    stateVersion: 47,         // Optimistic concurrency
    isActive: true,           // Lifecycle state
    lastActivity: timestamp,  // For auto-deactivation
    activityTimeout: 1800000  // 30min → auto-deactivate
}

Features:

• 🎭 Actor Lifecycle Management — Activate/deactivate based on activity
• 📊 Consistent Hash Partitioning — User ID → Partition (4 partitions)
• 💾 localStorage State Persistence — Actor state survives refresh
• 🔄 State Versioning — Optimistic concurrency control
• 🩺 Health Checks — Periodic liveness monitoring

Request routing with partition affinity:

PartitionRouter {
    partitions: Map(4),  // 4 logical partitions
    // Each partition tracks:
    {
        id: 0,
        load: 142,       // Operation counter
        lastSync: ts,    // Last sync timestamp
        healthy: true    // Health status
    }
}

Features:

• 🎯 Sticky Session Routing — Same user → same partition
• 📈 Load Tracking — Per-partition operation metrics
• 🩺 Health Monitoring — Partition failure detection
• ⚖️ Consistent Hashing — Deterministic partition assignment

Resilience pattern preventing cascade failures:

RequestBalancer {
    maxConcurrent: 5,        // Concurrent request limit
    maxQueueSize: 20,        // Backpressure limit
    maxFailures: 25,         // Circuit trip threshold
    circuitResetTime: 5000,  // Auto-heal timeout
    circuitBreakerOpen: false
}

Features:

• 🔌 Circuit Breaker — Opens after 25 failures, auto-heals in 5s
• ⚡ Priority Queue — Critical(10) → High(8) → Normal(5) → Low(1)
• 🔁 Request Deduplication — Prevents duplicate in-flight requests
• 🎚️ Adaptive Throttling — Max 5 concurrent, 5ms minimum delay
• 📊 Backpressure Handling — Drops lowest priority when queue full

Full-text search with relevance scoring:

SearchIndex {
    byWord: Map(),      // word → [{item, field, score}]
    byPrefix: Map(),    // prefix → Set(items)  
    bookmarks: []       // All indexed items
}

Features:

• 📇 Inverted Index — O(1) word lookups
• 🔤 Prefix Indexing — Autocomplete-style matching
• 📊 TF-IDF-like Scoring — Title(10) > Tag(8) > Category(6) > URL(4)
• 💨 LRU Result Cache — 50-entry query cache
• 🔀 Multi-word AND Search — All terms must match

Offline-capable with localStorage persistence:

// Actor state persisted to localStorage
localStorage.setItem(`state_p${partitionId}`, JSON.stringify({
    data: Object.fromEntries(localState),
    version: stateVersion,
    timestamp: Date.now()
}));

// Event queue persistence
localStorage.setItem('events', JSON.stringify(eventQueue));

// User identity persistence  
localStorage.setItem('user_id', uniqueUserId);
localStorage.setItem('instance_id', actorInstanceId);

Features:

• 💾 Actor State Persistence — Survives page refresh
• 📨 Event Queue Persistence — No lost messages
• 🆔 Identity Persistence — Consistent user tracking across sessions
• ⚡ Instant Hydration — Load from localStorage before backend fetch
• 🔄 Background Sync — Async reconciliation with server

Backend:

• 🔒 LockService — Google's distributed lock with 3s timeout
• ⚡ Optimistic Locking — Try-lock with immediate fail-fast
• 🔄 Automatic Retry — Frontend retries with backoff

Frontend:

• 📊 State Versioning — Conflict detection via version numbers
• 🔄 Optimistic UI — Instant feedback, rollback on failure
• 🆔 Collision-free IDs — timestamp + random generation

Cross-user synchronization with 1-2 second latency:

// Backend: Change notification flag
{
    timestamp: 1706961234567,
    userId: "kostas",
    action: "addBookmark"
}

// Frontend: 1-second polling
if (notification.timestamp > lastKnownChangeTime) {
    triggerFullSync();  // Priority 10 (highest)
}

Features:

• 📡 Change Notification Polling — 1-second intervals
• 👥 User Attribution — "John added bookmark" messages
• 🔢 Active User Count — Real-time presence indicator
• ⚡ Instant Propagation — <2s cross-user sync

🆓 $0/month — Replaced paid SaaS with enterprise-grade architecture on free tier